The RNA-Binding Protein QKI Controls Alternative Splicing in Vascular Cells

RESEARCH ARTICLE The RNA-binding protein QKI controls alternative splicing in vascular cells, producing an effective model for therapy Rachel Caines1, Amy Cochrane1, Sophia Kelaini1, Marta Vila-Gonzalez1, Chunbo Yang1, Magdalini Eleftheriadou1, Arya Moez1, Alan W. Stitt1, Lingfang Zeng2, David J. Grieve1 and Andriana Margariti1,*

ABSTRACT vessel (Carmeliet, 2000). VSMCs are not terminally differentiated Dysfunction of endothelial cells (ECs) and vascular smooth muscle as adult cells. In disease scenarios, cell plasticity can be evoked, cells (VSMCs) leads to ischaemia, the central pathology of with phenotypic changes to VSMCs potentiating the development cardiovascular disease. Stem cell technology will revolutionise of CVD leading to complications such as atherosclerotic lesion regenerative medicine, but a need remains to understand key rupture and postangioplasty restenosis (Owens et al., 2004; Davies mechanisms of vascular differentiation. RNA-binding proteins have et al., 2010; Hao et al., 2002; Moiseeva, 2001; Schatteman et al., emerged as novel post-transcriptional regulators of alternative splicing 1996; Belaguli et al., 1999; Bennett et al., 2016). and we have previously shown that the RNA-binding protein Quaking Thus far, treatment for vascular diseases has been largely (QKI) plays roles in EC differentiation. In this study, we decipher the preventative rather than replacing diseased tissue due to limited role of the alternative splicing isoform Quaking 6 (QKI-6) to induce availability of autologous tissue for transplantation. Cell VSMC differentiation from induced pluripotent stem cells (iPSCs). reprogramming is a powerful technique that has led to the PDGF-BB stimulation induced QKI-6, which bound to HDAC7 intron 1 generation of induced pluripotent stem cells (iPSCs) from adult via the QKI-binding motif, promoting HDAC7 splicing and iPS-VSMC somatic cells, which can be directed towards any cell type required differentiation. Overexpression of QKI-6 transcriptionally activated for therapy (Takahashi and Yamanaka, 2006). SM22 (also known as TAGLN), while QKI-6 knockdown diminished ECs and VSMCs have been successfully generated from iPSCs differentiation capability. VSMCs overexpressing QKI-6 demonstrated (iPS-ECs and iPS-SMCs, respectively) but many of the mechanisms greater contractile ability, and upon combination with iPS-ECs- of differentiation are still unknown, leading to limited efficiency of overexpressing the alternative splicing isoform Quaking 5 (QKI-5), derived differentiated cells (Margariti et al., 2012; Clayton et al., exhibited higher angiogenic potential in vivo than control cells 2015). Studying the underlying mechanisms of vascular cell alone. This study demonstrates that QKI-6 is critical for modulation of differentiation will allow the manipulation of vascular cells in the HDAC7 splicing, regulating phenotypically and functionally robust iPS- diseased state and provide novel targets for vascular therapy while VSMCs. These findings also highlight that the QKI isoforms hold key teaching us the underpinning molecular mechanisms of development. roles in alternative splicing, giving rise to cells which can be used in We have previously shown that embryonic stem cell differentiation vascular therapy or for disease modelling. into VSMCs requires signalling via histone deacetylase 7 (HDAC7), a class II histone deacetylase that is essential for tight regulation of gene This article has an associated First Person interview with the first expression (Zhang et al., 2010; Margariti et al., 2009; Dressel et al., author of the paper. 2001). Alternative splicing of HDAC7 plays a key role in SMC differentiation from pluripotent stem cells, while platelet-derived KEY WORDS: Vascular smooth muscle cell, Cellular reprogramming, growth factor-BB (PDGF-BB) is known to enhance VSMC Cell signalling, Stem cells, Revascularisation differentiation and regulate the balance of spliced to unspliced HDAC7 (Margariti et al., 2009; Zhou et al., 2011). HDAC7s INTRODUCTION modulates the serum response factor (SRF)–myocardin complex, Cardiovascular disease (CVD) is the leading cause of morbidity and which induces SMC differentiation from pluripotent stem cells mortality in the western world, and is characterised by progressive (Zhang et al., 2010; Margariti et al., 2009; Wang et al., 2004; Chen damage to and closure of blood vessels in key tissues (Roth et al., et al., 2002). Further elucidation of the exact splicing regulators of 2015). CVD is initiated by dysfunction of the vascular cells; the HDAC7 is required to allow full understanding of HDAC7-mediated endothelial cells (ECs) and vascular smooth muscle cells (VSMCs) VSMC differentiation, which will permit the generation of more (Park and Park, 2015). The endothelial interface between the vessel phenotypically mature VSMCs. and circulating blood is supported by VSMCs, playing essential RNA-binding proteins (RBPs) have emerged in recent years as roles in maintaining vascular tone and maturation of the blood important modulators of post-transcriptional regulation in the cell, through altering mRNA splicing, stability, localisation and efficiency of translation (Brinegar and Cooper, 2016). This leads to changes in 1Wellcome-Wolfson Institute for Experimental Medicine, Queen’s University Belfast, countless cellular process, with recent studies uncovering the roles of 97 Lisburn Road, Belfast BT9 7BL. 2Cardiovascular Division, King’s College London, London SE5 9NU, UK. RBPs in the maintenance of pluripotency and commitment to cell differentiation. For example, RNA-binding motif protein 3 (RBM3) *Author for correspondence ([email protected]) was found to be vital in osteoblast differentiation (Kim et al., 2018), A.M., 0000-0001-9303-8371 while heterogeneous nuclear ribonucleoprotein K (HNRNPK) has been found to regulate myoblast proliferation and differentiation (Xu

Received 23 January 2019; Accepted 10 July 2019 et al., 2018). Concurrently, dysregulation of RBP function is Journal of Cell Science

1 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276 implicated in a number of diseases such as diabetes and vascular immunohistochemistry, highlighting there was a robust dysfunction (Yang et al., 2018). differentiation towards VSMC fate (Fig. 1C,D). Our laboratory The RBP Quaking (QKI), a member of the signal transduction has recently reported that the RBP QKI-5 holds a key role in EC and activation of RNA (STAR) family of proteins, and its various differentiation, angiogenesis and neovascularisation (Cochrane isoforms have been demonstrated to be key in vascular development et al., 2017). Since the field of RBPs is fast emerging, and RBPs (Li et al., 2003; Noveroske et al., 2002). QKI transcription begins are now recognised as powerful, versatile regulatory units, which from one major start site but alternative splicing leads to the play pivotal roles in the regulation of cell differentiation (Guallar production of three protein isoforms: QKI-5, QKI-6 and QKI-7. and Wang, 2014), we hypothesised that QKI alternative splicing Each isoform has a unique C-terminal sequence, while sharing a may play a role in determining the developmental fate of the common RNA-binding sequence, or Quaking response element vascular cell. While QKI-5 expression did not increase above (QRE) within the body of the protein (Ebersole et al., 1996; Kondo control levels during SMC differentiation, the level of QKI-6 was et al., 1999; Galarneau and Richard, 2005). significantly induced over time at the mRNA and protein levels QKI was originally attributed as having a role in post-natal (Fig. 1B,C). It has been reported that myoblasts must initially myelination of the central nervous system, while subsequent express QKI-5 before being able to express any further QKI experimentation uncovered a role for QKI in blood vessel alternative splicing isoform due to the role of QKI-5 in promoting development. Mouse embryos with mutant QKI showed embryonic the accumulation and alternative splicing of the QKI gene (Fagg lethality at embryonic day (E)10–12.5 with only primitive vascular et al., 2017). In our study, we demonstrate that QKI-5 and -6 exhibit networks present due to impaired EC differentiation and an inability to similar levels of expression early in VSMC differentiation with recruit and differentiate mural cells (Noveroske et al., 2002). levels of QKI-6 then exceeding QKI-5 at later time points, QKI alternative splicing isoform (QKI-5) has recently been indicating that QKI-6 plays a role in VSMC differentiation identified by our laboratory to direct iPSC differentiation to an (Fig. 1C) (Fagg et al., 2017). In day 6 differentiated miPS-SMCs, endothelial lineage through direct binding and stabilisation of QKI-6 was expressed alongside the VSMC markers SM22 and STAT3, while iPS-ECs overexpressing QKI-5 enhanced repair in an calponin as shown by immunofluorescence staining (Fig. 1E; in vivo model of hind limb ischaemia (Cochrane et al., 2017). Fig. S1). It was also clearly demonstrated that QKI-6 expression is Notably, QKI has been shown to be strongly induced in vascular significantly influenced by the presence of PDGF-BB in the culture injury, altering the expression of myocardin, a key driver of VSMC medium. miPS-SMCs were differentiated for 24 h in the absence of differentiation (van der Veer et al., 2013). PDGF-BB before 10 or 25 ng/ml PDGF-BB was added to the Since vascular cells are capable of derivation from a common medium for 40 h. Both 10 and 25 ng/ml PDGF-BB led to a developmental progenitor (Yamashita et al., 2000), and QKI-5 is significant induction in QKI-6 mRNA expression in comparison to instrumental in EC differentiation from iPSCs (Cochrane et al., a non-treated control (Fig. 1F). 2017), we hypothesised that the alternative splicing isoform QKI-6 may play a role in facilitating key mechanisms of VSMC QKI-6 is implicated in the differentiation of SMCs from iPS differentiation from iPSCs. Since QKI-6 is co-localised with cells spliceosomal proteins in the nucleus, it may hold a pivotal role in Further experiments were conducted to evaluate the role of QKI-6 in regulating alternative splicing controlling VSMC differentiation. the differentiation of VSMCs. QKI-6 was overexpressed by In this study, we provide robust evidence that QKI-6 is highly lentiviral gene transfer (Ex-QKI-6) in day 4 differentiated miPS- induced during VSMC differentiation from iPSCs, stimulated by SMCs for 48 h, which enhanced VSMC marker expression such PDGF-BB, while directly regulating the splicing of HDAC7 and as calponin, myosin heavy chain (MHC; herein, MYH11) and SMA driving iPS-SMC differentiation. Through exploring the key at the mRNA level in comparison to the empty vector control miPS- mechanisms of QKI-5 and -6 in mediating alternative splicing SMCs (Ex-mCherry) (Fig. 2A). This data was confirmed in VSMCs programmes during vascular cell differentiation, we have increased overexpressing QKI-6 on day 1 of differentiation for 48 h, the efficiency of in vivo angiogenesis and created a combination cell indicating QKI-6 is capable of inducing SMC differentiation from therapy of QKI-5-derived ECs and QKI-6-derived VSMCs. miPSCs (Fig. S2). Moreover, the induction of QKI-6, calponin and SM22 (also known as TAGLN) were confirmed at the protein level RESULTS through western blotting and immunofluorescence staining QKI-6 is induced during VSMC differentiation mediated by (Fig. 2B,C; western blot quantification can be found in Fig. S3A). PBGF-BB We previously reported that the transcription factor known to play a Mouse iPSCs seeded on collagen IV and cultured in differentiation key role in angiogenesis, STAT3, is regulated by QKI-5 during EC medium (DM) supplemented with 25 ng/ml of PDGF-BB (hereafter differentiation. Interestingly, STAT3 was significantly reduced on known as DM+P), differentiated towards a typical VSMC QKI-6 overexpression indicating that VSMC differentiation is morphology (Fig. 1A, top panels) and expressed characteristic preferentially taking place (Fig. 2A). Quantification of cells VSMC markers such as α-smooth muscle actin (SMA, also known successfully expressing the mCherry control or QKI-6 as ACTA2) and calponin (Fig. 1A, bottom panels). In mouse iPS- overexpression vectors alongside the smooth muscle cell marker derived VSMCs (miPS-SMCs), SMA was significantly induced at SM22 indicated an increase in VSMC differentiation efficiency the mRNA level from day 4 of differentiation (Fig. 1B) with from 73% to 92% (Fig. 2C). Importantly, QKI-6 overexpression on increasing calponin and SM22 expression demonstrated at the day 3 of differentiation for 48 h also led to transcriptional activation protein level (Fig. 1C). of SM22 as shown by luciferase assay (Fig. 2D), suggesting QKI-6 Since VSMCs and ECs have the developmental capability to enhances miPS-SMC differentiation via transcriptional activation of derive from a common Flk-1+ progenitor cell (Yamashita et al., VSMC markers such as SM22. 2000), it was important to assess possible co-differentiation of ECs. In contrast, when QKI expression was knocked down through No CD144 (also known as CDH5) or von Willebrand factor (vWF) shRNA transfection (shQKI) on day 3 of miPS-SMC differentiation protein was present, as demonstrated by western blotting and for 72 h, expression of the VSMC markers SMA and SM22 were Journal of Cell Science

2 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 1. iPSC differentiation towards SMCs. (A) iPSCs were seeded on collagen IV and cultured in DM+P to induce SMC differentiation. Morphology of iPSCs (top left panel) and their SMC differentiated counterparts (top right panel) are shown in bright-field microscopy images, with immunofluorescence staining showing the SMC markers α-SMA (bottom left panel) and calponin (bottom right panel). Scale bar: 50 µm. (B) The SMC marker SMA was induced in a time-dependent manner during SMC differentiation from miPSCs in parallel with the RBP QKI-6 at the mRNA level. (C) Time point protein analysis shows induction of the SMC markers calponin and SM22 in parallel with QKI-6 during miPS-SMC differentiation. The EC marker CD144 and QKI-5, are not induced. (D) Immunofluorescence staining of day 6 miPS-SMCs reveals QKI-6 staining throughout the culture while no CD144 or vWF expression could be observed. Scale bar: 50 µm. (E) Immunofluorescence staining shows co-expression of QKI-6 and SMC markers SM22 and calponin in day 6 miPS-SMCs. Scale bar: 50 µm. (F) qRT-PCR shows that supplemental PDGF-BB is capable of significantly inducing QKI-6 expression at the mRNA level. Data are mean±s.e.m. (n=3). *P<0.05, **P<0.01, ***P<0.001.

significantly reduced at the mRNA and protein level in comparison to removing three stop codons and allowing transcription to begin what was found in cells transfected with a non-targeting control from an alternative splice site. This process leads to a 22 amino acid shRNA (shNT) (Fig. 2E,F). These findings indicate that QKI-6 has a longer HDAC7 spliced isoform (HDAC7s). The expression of role in VSMC differentiation from iPSCs and that it could be used to the shorter, unspliced HDAC7 (HDAC7u) readily leads to the enhance the efficacy of this process. degradation of myocyte enhancer factor 2c (MEF2c) via the proteasome, suppressing VSMC differentiation. By contrast, QKI-6 directs signalling and promotes splicing of HDAC7 the HDAC7s variant localises to the nucleus, where it binds to We have previously reported that PDGF-BB regulates the SRF and recruits myocardin, which is known to be a key step in transcription and subsequent splicing of HDAC7, which has a SMC differentiation (Margariti et al., 2009). In this study, we have role in directing the differentiation of embryonic stem cells (ESCs) found that PDGF-BB regulates QKI-6 expression and promotes to VSMCs via the SRF–myocardin complex (Margariti et al., 2009). VSMC differentiation. The next question was whether QKI-6, as an

Splicing of HDAC7 leads to the exclusion of a 57-base-pair intron, RBP, could regulate HDAC7 signalling and splicing. Journal of Cell Science

3 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 2. QKI-6 is implicated in and required for SMC differentiation from iPSCs, and drives transcriptional activation of SM22. iPSCs were seeded on collagen IV and cultured in DM+P for 4 days. QKI-6 was overexpressed by lentiviral transduction and cells were harvested 48 h later. (A) QKI-6-mediated induction of SMC markers calponin, MHC and SMA was shown at the mRNA level by qRT-PCR. STAT3 mRNA was significantly reduced. (B) QKI-6 overexpression was shown to induce expression of calponin and SM22 at the protein level in comparison to an mCherry control. (C) Immunohistochemistry was employed to assess those cells expressing SMC markers upon overexpression of QKI-6. Quantification in relation to an empty vector control indicated VSMC differentiation efficiency increased from 73% to 92% upon QKI-6 overexpression. Low magnification imaging in the right-most panels demonstrates a significant increase in SM22 expression upon QKI-6 overexpression. Scale bars: 50 µm. (D) miPS-SMCs were transduced with a luciferase reporter plasmid encoding for the SM22 promoter region and co- transfected with Ex-mCherry or Ex-QKI-6 on day 3 of differentiation. The transcriptional activity of SM22 was found to be significantly increased upon overexpression of QKI-6 via a readout of the relative luciferase units (RLUs), indicating its role in driving transcriptional activation of SMC differentiation. (E) iPSCs were seeded on collagen IV and cultured in DM+P for 3 days when QKI was knocked down by means of lentiviral shRNA (shQKI). Analysis at the mRNA level, 72 h after infection, shows a significant reduction in expression of SMC markers SMA and SM22 after knockdown versus a non-targeting control (NT). (F) The effect of QKI knockdown on the reduction of QKI-6, SM22 and calponin were confirmed by western blotting. Data are mean±s.e.m. (n=3). *P<0.05, **P<0.01.

QKI-6 overexpression on day 1 of differentiation for 48 h led to a factors, such as polypyrimidine track-binding protein (PTB; also significant induction of HDAC7, SRF and myocardin expression at known as PTBP1) during muscle cell differentiation (Hall et al., the mRNA (Fig. 3A) and protein level (Fig. 3B; western blot 2013). Hence, a screen was carried out in early differentiating cells for quantification can be found in Fig. S3B), while QKI knockdown at the induction of a range of alternative splicing factors, including QKI- day 3 of differentiation for 72 h significantly reduced SRF 5 and QKI-6. A number of genes were significantly downregulated expression (Fig. 3C). Induction of HDAC7 splicing during between day 2 and day 4 of miPS-SMC differentiation (Fig. 3E). differentiation of miPSCs to SMCs was confirmed by qRT-PCR Notably, Hnrnpa2b1, known to play a role in SMC differentiation (Fig. 3D; splicing specific primer detection detailed in Fig. S4), from ESCs was downregulated. Hnrnpa2b1 has been shown to peak but studies to date do not address the exact splicing regulators between day 0 and 3 of SMC differentiation followed by a decline in of HDAC7. expression, which could be the decrease we are observing in this study It is known that QKI can have fundamental roles in splicing alone (Wang et al., 2012). Hnrnpf is known to be repressed by QKI-6 in

(van der Veer et al., 2013), or in combination with a range of splicing oligodendrocytes and a similar response appears to be occurring Journal of Cell Science

4 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 3. QKI-6 is implicated in HDAC7 signalling and promotes its splicing. (A,B) QKI-6 was overexpressed by plasmid transfection on day 1 of SMC differentiation and cells were harvested 48 h later. QKI-6 overexpression was shown to significantly induce expression of HDAC7 along with members of the downstream SMC signalling pathway (SRF and myocardin) at the mRNA and protein level. (C) Knockdown of QKI on day 3 of differentiation for 72 h suppressed SRF expression significantly, as shown by qRT- PCR. (D) Levels of HDAC7 and its splicing were observed throughout miPS-SMC differentiation. HDAC7 splicing increased, showing a significant increase in expression at day 6 of differentiation in comparison to splicing at day 2 miPS- SMCs as revealed by RT-qPCR. Levels of total HDAC7 were not significantly changed. (E) miPS-SMCs between day 2 and day 4 of differentiation were screened for the induction of a range of relevant splicing factors, including QKI-5 and QKI-6. qRT-PCR revealed only QKI-6 was significantly induced at day 4 of differentiation in comparison to day 2 differentiated cells. Data are mean±s.e.m. (n=3). *P<0.05, **P<0.01, ***P<0.001.

in differentiating miPS-SMCs (Mandler et al., 2014). Of the Srsf differentiation, immediately prior to maximum levels of HDAC7 isoforms, Srsf1 has been investigated for its role in neointimal expression being reached within the cells (Fig. 3D). HDAC7 splicing hyperplasia, driving excessive VSMC proliferation. As the cells increases during miPS-SMC differentiation (Fig. 3D), explaining why redifferentiate to a contractile state, Srsf1 levels are reduced (Xie et al., baseline levels of spliced HDAC7 vary in Fig. 4A and C. 2017). As the miPS-SMCs differentiate and become more contractile, To verify that QKI-6 acts upstream of HDAC7 to impact on Srsf1 would be expected to decrease as observed. Interestingly, QKI-6 VSMC differentiation from miPS-SMCs, HDAC7 was knocked was the only transcript found to be significantly induced during this down on day 5 of differentiation followed by QKI-6 overexpression process (Fig. 3E), therefore we sought to identify the association on day 6. Cells were harvested on day 8 of miPS-SMC between QKI-6 and HDAC7 splicing. differentiation. Upon QKI-6 overexpression, a significant increase Overexpression of QKI-6 was shown to enhance endogenous of HDAC7, its splicing and calponin expression were observed, expression of both HDAC7 isoforms as shown by conventional PCR whereas following knockdown of HDAC7, HDAC7 and calponin (Fig. 4A). cDNA was also subjected to qRT-PCR and a significant expression were significantly reduced. When HDAC7 was knocked induction of HDAC7 splicing was observed in QKI-6 overexpressing down, the effects of QKI-6 overexpression on HDAC7 splicing and cells in comparison to mCherry-transfected control cells (Fig. 4B). calponin expression were ablated (Fig. 4E). Similarly, upon knockdown of QKI, splicing of HDAC7 was reduced As all QKI isoforms share the same KH RNA-binding domain in comparison to cells transfected with non-targeting control and a and only differ at their C-terminus, we wished to assess that the significant drop in HDAC7 splicing was determined by qRT-PCR observed effect on HDAC7 splicing was in fact QKI-6 specific. (Fig. 4C,D). In these experiments, overexpression of QKI-6 was QKI-5 was overexpressed in day 3 differentiating miPS-SMCs carried out on day 1 of differentiation in order to manipulate early cultured in DM+P and harvested 48 h after transfection. QKI-5 was splicing events, while knockdown was employed on day 3 of unable to induce the expression of VSMC markers calponin or Journal of Cell Science

5 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 4. QKI-6 acts upstream of HDAC7 and directs its splicing. (A) Conventional PCR was carried out from cDNA of cell lysates overexpressing QKI-6 with primers designed to detect levels of HDAC7 splicing (HDAC7s; 314 base pairs) versus unspliced HDAC7 (HDAC7u; 371 base pairs). Overexpression of QKI-6 leads to increases in both spliced and unspliced HDAC7. (B) Effect of QKI-6 overexpression on the level of spliced HDAC7 mRNA was quantified by qRT-PCR. (C,D) Knockdown of QKI on day 3 of SMC differentiation for 72 h led to a decrease in levels of spliced HDAC7 (C); this is quantified by qRT-PCR (D). (E) To verify the QKI-6- and HDAC7-mediated mechanism of SMC differentiation and to confirm that QKI-6 lies upstream of HDAC7, HDAC7 was knocked down in day 5 miPS-SMCs, followed by QKI-6 overexpression on day 6 alongside relevant combinations of Ex-mCherry and NT controls. Cells were harvested after 48 h and qRT-PCR was carried out. Upon knockdown of HDAC7, QKI-6 overexpression was no longer able to exert its effect on inducing HDAC7, HDAC7 splicing or calponin expression. (F) Overexpression of QKI-5 in day 3 differentiating miPS-SMCs led to a significant reduction in HDAC7 splicing, but was unable to induce SMC marker expression as previously shown by QKI-6 overexpression. Differentiation was also not shown to be directed to an EC lineage upon overexpression of QKI-5 since endothelial cell markers were not observed. Data are mean±s.e.m. (n=3). *P<0.05, **P<0.01, ***P<0.001.

SM22 while levels of HDAC7 splicing were significantly reduced. occurs. To identify whether QKI could bind directly to this site to In addition, endothelial cell markers were not observed (Fig. 4F). modulate splicing, primers incorporating the binding site were These findings provide strong evidence that QKI-6 specifically acts designed (Fig. 5B) and RNA immunoprecipitation was carried out. upstream of HDAC7 and QKI-6 must act through the splicing of miPS-SMCs transduced to overexpress QKI-6 on day 4 of HDAC7 to subsequently drive SMC differentiation from iPSCs. QKI- differentiation for 48 h underwent pulldown with IgG control or 5 and QKI-6 may also compete for binding to the HDAC7 transcript anti-QKI-6 antibody. RNA was extracted from the resultant RBP– and recruit different machinery to impact mRNA turnover and splicing. mRNA complex and conventional PCR was undertaken. QKI-6 pulldown showed strong amplification of the HDAC7 intron 1 QKI-6 binds directly to HDAC7 mRNA binding site. Fig. 5C is a representative image of three independent Precisely how QKI-6 interacts with HDAC7 to direct its splicing experiments. Relative band intensity was quantified across all three during miPS-SMC differentiation remains unknown. We therefore experiments and showed significantly increased binding of QKI-6 screened the HDAC7 mRNA sequence for the conserved QKI- to HDAC7 intron 1 versus IgG control (Fig. 5C,D). Interestingly, binding motif ‘ACUAAC’ using the RBPmap software (Galarneau the conserved QKI-binding motif was also identified in intron 1 of and Richard, 2005; Paz et al., 2014). A conserved QKI-binding site the human HDAC7-212 isoform. An RNA-binding assay was within intron 1 of HDAC7 at the genomic location chr15:97842906 carried out in 293T cells overexpressing QKI-6 for 48 h and QKI was identified (Fig. 5A), intron 1 being where HDAC7 splicing was again seen to directly bind to this location (Fig. S5). Journal of Cell Science

6 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 5. QKI-6 binds directly to HDAC7 intron 1. (A) RBPmap software revealed a conserved QKI- binding motif that was predicted to align to intron 1 of HDAC7. (B) Specific primers were designed to incorporate this binding motif and amplify a region of 249 base pairs. (C) RNA immunoprecipitation was carried out on miPS- SMCs transduced to overexpress QKI-6 on day 4 of differentiation for 48 h. Pulldown by QKI-6 antibody showed a direct binding to HDAC7 intron 1 through conventional PCR of the resultant cDNA. Image shown is representative of three independent experiments. (D) Relative band intensity of IgG versus QKI-6 pulldown was quantified using ImageJ. Data are mean±s.e.m. (n=3). ***P<0.001.

Combining QKI-5-derived ECs with QKI-6-derived SMCs In vivo, cells underwent QKI overexpression as described above increases angiogenic potential in vitro and in vivo and Matrigel plug assays on the flanks of C57BL6 mice demonstrated To understand the functional role and potential applications of QKI- that both control (mCherry) and QKI-derived miPS-SMCs and derived vascular cells, a series of in vitro and in vivo functional assays miPS-ECs produced a significantly greater number of vascular werecarriedout.Day3miPS-SMCstransfectedwithQKI-6for48h structures per image than the PBS (vehicle) control. Remarkably, exhibited enhanced contractile capabilities compared to controls, there was a significant increase in number of vascular-like indicating more functional cells (Fig. 6A). In order to study the tube- structures present when QKI-5- and QKI-6-derived miPS-ECs and forming capacity of the cells, ECs and SMCs differentiated from iPSCs miPS-SMCs were combined versus mCherry-derived miPS-vascular were transfected with a QKI-5 or QKI-6 overexpression vector, cells (Fig. 6C,D). Immunofluorescence staining demonstrated respectively, on day 3 of differentiation, and vascular tube formation incorporation of injected cells into vessel-like structures, assessed after 48 h of overexpression. QKI-derived vascular cells had successfully expressing CD144 and SM22 in vivo (Fig. 6E; Fig. S6). significantly increased segment tube length and branch number in comparison to controls (Fig. 6B). Persistence of QKI-derived vascular DISCUSSION structures was not assessed in comparison to controls, but due to the VSMCs are highly heterogeneous cells and are derived from added stability VSMCs provide to vessels, we would hypothesise that numerous developmental locations (Majesky, 2007). The origin of these structures would have increased longevity in culture and would VSMCs has been shown to impact greatly on the development of be less likely to degrade as quickly as controls or ECs alone. vascular diseases such as atherosclerosis (Bennett et al., 2016; Journal of Cell Science

7 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 6. QKI-6-derived SMCs exhibit greater contractile ability and, on combination with QKI5-derived ECs, have an increased angiogenic potential in vitro and in vivo. (A) Day 3 miPS-SMCs were transfected with Ex-QKI6 or Ex-mCherry and, after 48 h, were stimulated with KCl. Live imaging was carried out from time of stimulation for 15 min. Transfected cell size was measured at time zero and after 15 min. The percentage change in cell size was calculated. QKI-6-overexpressing cells showed a significantly greater decrease in cell size, indicating their contraction in comparison to mCherry controls. (B) Day 3 miPS-ECs and SMCs were transduced to overexpress Ex-mCherry for both cell types, Ex-QKI-5 for ECs or Ex-QKI-6 for SMCs. After 48 h, cells were subjected to an in vitro Matrigel assay. Combined QKI-5 miPS-ECs and QKI-6 miPS-SMCs produced longer vessels as shown by an increase in the total master segment length (top right panel) and significantly increased levels of vessel branching (bottom right panel). Representative images of mCherry- and QKI-derived vascular cells are shown on the left. Scale bars: 75 µm. (C,D) Cells subjected to the same treatments as in B were also used for an in vivo Matrigel assay. Cells were mixed with Matrigel and injected subcutaneously. The total number of vessel-like structures were quantified after harvest 1 week later. Vessel numbers were calculated per image and averaged for each plug. Plugs containing mCherry- and QKI-induced vascular cells showed significantly increased vessel numbers versus PBS control and there was a significantly increased vessel number in QKI-derived cell plugs versus mCherry-derived cell plugs. Scale bars: 25 µm. (E) In vivo Matrigel plugs were subjected to immunohistochemistry to show direct incorporation of injected cells into the vasculature. Injected cells were labelled with Vybrant (red) and were seen to colocalise

with CD144 and SM22 staining within the QKI-derived plugs. Scale bars: 75 µm. Data are mean±s.e.m. (n=3) *P<0.05, **P<0.01, ***P<0.001. Journal of Cell Science

8 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

DeBakey and Glaeser, 2000). Here, VSMCs can act as protective differentiation, tightly orchestrated alternative splicing events occur, stabilisers of the atherosclerotic plaque, while others dedifferentiate, which are highly sensitive to the cellular environment and other proliferate and induce apoptosis in response to the local environment stimuli. Previously and in this study, we have demonstrated how the (Libby et al., 2011; Gomez and Owens, 2012). This can lead to presence of VEGF induces EC fate and PDGF-BB directs VSMC plaque instability, rupture and further complications after therapeutic fate to differentiating iPSCs. These culture conditions also implicate intervention such as in-stent restenosis (Qian et al., 2007; Alfonso QKI splicing events inducing the specific splicing isoform required et al., 2014). Other conditions that have a microvascular component, for precise vascular cell development. This demonstrates one of the such as diabetes, are known to be potentiated by VSMC regression multiple regulatory layers of cell differentiation and lineage and loss (Montero et al., 2013; Beltramo and Porta, 2013). commitment (Wang et al., 2008; Fu and Ares, 2014). By understanding the developmental programming of VSMCs QKI-6 is capable of driving HDAC7 splicing and SMC gene through investigation of transcriptional and epigenetic mechanisms expression while its absence eradicated this ability. Adding further of differentiation from iPSCs, we can identify novel therapeutic PDGF-BB during differentiation was found to be integral in targets and produce innovative cell therapies for ischaemic disease. inducing the expression of QKI-6, linking the growth factor to the Alternative splicing of mRNA allows the production of several previous finding that PDGF-BB regulates levels of spliced HDAC7 protein products from one gene locus (Yabas et al., 2015). This within VSMCs. QKI-6 regulated HDAC7 splicing by directly largely impacts on the scale of genetic diversity and can occur in a binding to the splicing region of HDAC7 in intron 1. Furthermore, cell- or tissue-specific manner in response to environmental cues. QKI-6 was unable to direct splicing or SMC differentiation to the Alternative splicing allows for fine-tuning of complex cellular same extent in the absence of HDAC7 expression, emphasising the responses such as cell differentiation in health and disease (Nilsen important role of the mechanism of QKI-6-mediated HDAC7 and Graveley, 2010; Black, 2003; Modrek et al., 2001). In 2002, splicing in VSMC differentiation. Noveroske et al., and in 2003 Li et al. detailed lethal blood vessel Neff et al. have previously discussed how VSMCs are often defects and loss of vessel integrity in mouse embryos null or forgotten in the generation of tissue engineered blood vessels. deficient for the RBP QKI, respectively. These effects were due to Although ECs are the main functional unit of the blood vessel, SMCs loss of the QKI-5 splice site, resulting in an inability to remodel the contribute to a much larger portion of the vessel wall and are crucial yolk sac vasculature, a combination of poor EC development and a in development and remodelling of the vasculature along with deficiency of VSMC recruitment and maturation (Li et al., 2003; maintaining vascular structure and tone in response to physiological Noveroske et al., 2002). This demonstrates the importance of QKI cues (Neff et al., 2011; Benjamin et al., 1998; Lilly, 2014). Therefore, alternative splicing in successful embryonic development. In recent it is logical that using ECs alongside VSMCs when studying vascular years, the mechanistic details of how QKI regulates splicing during development will lead to improved understanding of vascular differentiation has begun to be investigated. Fagg et al. detailed physiology. On combination of QKI-derived miPS-ECs and miPS- cross-regulation of the QKI isoforms in directing differentiation of SMCs, we saw increased angiogenic potential in vitro and in vivo,in myoblasts, while van der Veer et al. have shown that QKI is capable single and co-culture conditions. This demonstrates a basis for using of regulating myocardin expression after arterial damage of adult the QKI splicing isoforms in vascular therapy. SMCs (van der Veer et al., 2013; Fagg et al., 2017). Owing to the In conclusion, this study elucidates a novel role for the RBP QKI-6 role of QKI in adult and embryonic vascular cells, we hypothesised and provides further insight into the splicing mechanisms of HDAC7 that QKI could be involved in the differential fate of vascular cells. during SMC differentiation from miPSCs. We have demonstrated iPSC technology has revolutionised the future of regenerative mechanisms of vascular cell differentiation from iPSCs in which QKI medicine therapy and disease modelling, but to maximise the alternative splicing in response to lineage-specific growth factors is impact of this for CVD therapy, a thorough understanding is crucial. QKI-5 binds to the 3′-UTR of STAT3, leading to its required of the post-transcriptional mechanisms of iPSC stabilisation and phosphorylation, resulting in VEGFR2 activation differentiation. Here, we show how alternative splicing of the and VE-cadherin stabilisation. This is mediated upstream by VEGF- RBP QKI during differentiation of iPSCs to vascular cells results in induced expression of ETS-1, promoting QKI-5 function (Cochrane a cell fate decision directed by the choice of key growth factors et al., 2017). Conversely, PDGF-BB leads to the induction of QKI-6 supplied to the cells, which direct cell differentiation towards a and iPSC differentiation towards a VSMC fate. QKI-6 binds directly specific vascular cell lineage such as VSMCs. to intron 1 of HDAC7, causing a 57-base-pair intron to be excised, We recently reported how during EC differentiation from iPSCs, and a splicing event to occur. β-catenin is then capable of inducing under the control of vascular endothelial growth factor (VEGF), QKI- SMC proliferation while spliced HDAC7, in conjunction with SRF 5 acted to stabilise VE-cadherin expression and VEGFR2 and myocardin, activate and drive expression of SMC genes such as transcriptional activation, increasing the efficiency of differentiation SM22 and calponin, therefore driving a VSMC fate (Zhang et al., and leading to increased neovascularisation and angiogenesis 2010; Margariti et al., 2009; Zhou et al., 2011; Yang et al., 2016) (Cochrane et al., 2017). On addition of PDGF-BB during (Fig. 7). We envisage that these iPS-derived vascular cells could be differentiation of miPS-SMCs, QKI-6 is selectively induced and employed for tissue engineering, disease modelling and drug directly impacts on alternative splicing of HDAC7, resulting in a screening in a patient-specific manner prior to administration to the greater SMC differentiation potential. We show that in early diseased patient. This will lead to more effective, efficient and differentiation of SMCs, QKI-5 is expressed but this is maintained personalised cell sources for cardiovascular cell therapies. at low levels while QKI-6 continues to be induced. This demonstrates the multi-layered complexity of iPSC differentiation and how MATERIALS AND METHODS alternative splicing can be harnessed to alter cell fate decisions. Materials RBPs are part of highly complex systems and require tight Cell culture medium, serum and cell culture supplements were purchased regulation of their cellular localisation, expression levels, translation from the ATCC, Millipore and Thermo Fisher Scientific. efficiencies and splicing events for normal cell development and Antibodies against calponin [ab46794; 1:2000 for western blotting (WB), stability (Baralle and Giudice, 2017). Particularly during cell 1:500 for immunocytochemistry (ICC)], SM22 (ab14106; 1:1000, WB; Journal of Cell Science

9 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Fig. 7. Molecular mechanisms of QKI-directed vascular cell differentiation. Lineage-specific growth factors dictate the expression of the QKI alternative splicing isoforms QKI-5 and QKI-6. VEGF induces the transcription factor ETS-1, promoting QKI-5 binding to STAT3, STAT3 stabilisation and phosphorylation, VEGFR2 activation and VE-cadherin stabilisation. PDGF-BB induces QKI-6 binding to intron 1 of HDAC7, leading to its splicing and promotion of the SRF– myocardin SMC differentiation cascade. Combining QKI-derived vascular cells could be used in numerous applications such as tissue engineering, drug testing and disease modelling.

1:200, ICC), QKI-6 (ab9906, 1:500, ICC), QKI-5 (ab9904; 1:500, WB; 2-mercaptoethanol (Life Technologies #31350-010). The cells were 1:200, ICC), GAPDH (ab125247; 1:1000, WB) were purchased from passaged every 2 days at a ratio of 1:6. Cells were not used after passage Abcam. Antibodies against SRF (sc-335, 1:1000, WB), myocardin (H-300; 60. Differentiation of iPSCs towards vascular cells was induced by seeding 1:1000, WB) and vWF (C-20; 1:50, ICC) antibodies were obtained from the cells on type IV mouse collagen (5 μg/ml; R&D Systems #3410-010-01) Santa Cruz Biotechnology. A second QKI-6 antibody (N182/17; 1:1000, and cells were cultured in DM containing α-MEM (Life Technologies WB) was purchased from Neuromab. CD144 (STJ96234, 1:200k ICC; #32571036) supplemented with 10% FBS (Invitrogen #10270106), 1:1000, WB) and HDAC7 (STJ93484, 1:1000, WB) antibodies were 0.05 mM 2-mercaptoethanol, PenStrep, supplemented with 25 ng/ml purchased from St. John’s Laboratory, London, UK. Cy3-conjugated PDGF (Thermo Fisher Scientific #PMG0045) for SMC differentiation or α-smooth muscle actin (C6198, 1:100, ICC) was obtained from Sigma. 25 ng/ml VEGF (Thermo Fisher Scientific #PMG0111) for EC The secondary antibodies for immunostaining were anti-mouse-IgG and differentiation for the time points indicated. rabbit-IgG conjugated to Alexa Fluor 568, and anti-mouse-IgG and rabbit- For PDGF-BB stimulation experiments, miPSCs were subjected to IgG conjugated to Alexa Fluor 488 purchased from Thermo Fisher differentiation in the absence of PDGF-BB for 24 h. The cells were then Scientific and used at a 1:200 dilution. The secondary antibodies for western serum starved overnight and either 10 or 25 ng/ml PDGF-BB was added to blotting were purchased from Bio-Rad and used at a 1:3000 dilution. the media for 40 h. Cells were routinely tested for mycoplasma Control shNT (sc-62166) and shQKI (sc-106468) plasmids were purchased contamination fortnightly. from Santa Cruz Biotechnology and shHDAC7 from Sigma (NM_019572). 293T (ATCC #HEK293T) cells were cultured in DMEM (Thermo Fisher The plasmids for expression of QKI isoforms 5 (217EX-T4215-Lv224) and Scientific #10566016), Penstrep with 10% FBS. 6 (217EX-H2552-Lv224) were designed and purchased from Genecopoeia. Reverse transcription of RNA and qRT-PCR Cell culture and differentiation Total RNA was extracted using the RNeasy Mini Kit (Qiagen #74104) and Mouse iPSCs (miPSCs) were generated as previously described (Kelaini reverse transcribed by using a High Capacity cDNA Reverse Transcription et al., 2014; Di Bernardini et al., 2014). Mouse iPSCs were cultured on Kit (Thermo Fisher Scientific #4368814) with 10 µg of RNase inhibitor gelatin (PBS; Life Technologies #10010056 containing 0.02% of gelatin (Thermo Fisher Scientific #N8080119) according to manufacturer’s from bovine skin; Sigma #G1393) in DMEM (ATCC #30-2002) protocol. Resultant cDNA was diluted to a concentration of 10 ng/µl. supplemented with 10% fetal bovine serum (FBS) (Embryomax; Relative gene expression was determined by quantitative real-time PCR Millipore #ES-009-B), 100 IU/ml penicillin and 100 μg/ml streptomycin (qRT-PCR), with the SYBR Green Master Mix (Life Technologies #4368702) (PenStrep) (Thermo Fisher Scientific #10378-016), 10 ng/ml recombinant and primers as detailed in Table S1. GAPDH served as the endogenous human leukaemia inhibitory factor (LIF) (Millipore #LIF1010) and 0.1 mM control. The gene was considered undetectable beyond 35 cycles. Journal of Cell Science

10 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Lentivirus generation and infections normalised using 2× Laemmli sample buffer (Sigma #S3401) before 20– Lentivirus was generated in 293T (ATCC #HEK293T) cells passaged to 60– 50 µg of lysate was applied to SDS-PAGE and transferred to Amersham 70% confluency. 4.5 µg of relevant plasmid was combined with the Hybond P 0.45 PVDF blotting membrane (GE Healthcare #10600023). The packaging genes; 0.9 µg pCMV-dR8.2 (Addgene #8455; deposited by Bob bound primary antibodies were detected using horseradish peroxidase Weinburg) and 3.6 µg pCMV-VSV-G (Addgene #8454; deposited by Bob (HRP)-conjugated secondary antibodies (Bio-Rad #1706516, #1706515) Weinburg) with 27 µl EndoFectin Max (Genecopoeia #EF013) and 300 µl and Clarity Western ECL Substrate (Bio-Rad #1705061). Opti-MEM 1 Reduced Serum Medium (Thermo Fisher Scientific #31985070). This was incubated for 10 min at room temperature. Cells Indirect immunofluorescence cell staining assay were washed once with pure Opti-MEM which was followed by addition of Cell fixation was carried out with 4% paraformaldehyde and cells were 5 ml 2% FBS-Opti-MEM to the flasks. The transfection solution was added permeabilised with 0.1% Triton X-100 inPBS. Cultures were blocked in 5% dropwise to the cells and flasks incubated overnight at 37°C. Medium was goat or donkey serum, followed by incubation with the desired primary changed next morning to complete 293T medium (DMEM, Penstrep, 10% antibody. Subsequent washes were followed by secondary antibody FBS). After 48 h, the first round of virus was collected and passed through a incubation and counterstaining with DAPI (Life Technologies #D1306) 0.45 µm filter, removing any cell debris. Medium was replaced for another before mounting with Vectashield (Vector Laboratories #H-1000). Cells were 48 h when harvest process was repeated. imaged by fluorescence (Dmi8, Leica) and confocal microscopy (SP8, Leica). miPSC-derived vascular cells were allowed to differentiate for the indicated amount of time before lentiviral transduction. Equal volumes of control or Luciferase reporter assay overexpression/knockdown virus in DM were added to the cells with an On day 3 of miPS-SMC differentiation, cells were transfected using FuGENE additional 10 µg/ml of Polybrene (Santa Cruz Biotechnology #134220). 6 (Promega, # E2691) with Ex-mCherry or Ex-QKI-6 vector and the SM22 Medium was changed the next morning and cells were harvested 48 h later in promoter (pGL3-SM22). pGL3-Luc Renilla wasusedasaninternalcontrol. overexpression scenarios and 72 h after infection for knockdown. At 48 h after transfection, luciferase and Renilla luciferase activity was detected (Promega, #E1501; #S2001). Relative luciferase Units (RLU) were Gene overexpression by plasmid transfection defined as the ratio of luciferase activity to Renilla luciferase activity. Plasmids were transformed by combination with competent E. coli (Promega #JM109), plated on agar (Sigma, #L2897) with suitable RNA-binding assay resistance antibiotic added (Sigma, Ampicillin #BP785) and grown miPS-SMCs and 293T cells (ATCC #HEK293T) overexpressing QKI-6 were overnight. Plasmids from colonies picked and amplified the next day were harvested at 48 h post transfection and subjected to RBP immunoprecipitation purified using the QIAprep Spin Miniprep Kit (Qiagen #27106). DNA using the Magna RIP kit (Millipore, #17-700) as per the manufacturer’s concentration and purity was measured by the Nanodrop spectrophotometer. protocol. A QKI-6 specific antibody (Abcam, #ab9906) and rabbit IgG miPSCs were differentiated for the time points indicated before (included in kit) were used. Purified RNA was subjected to RT-PCR using transfection with 500 ng–2 µg of relevant plasmid. 3 µl of EndoFectin specific primers for the predicted HDAC7 intron 1 binding site. Max (Genecopoeia #EF013) was incubated per 1 µg of plasmid in 50 µl Opti-MEM per reaction for 10 min at room temperature. A reduced volume Contraction assay of Opti-MEM-2%FBS was added to the culture dishes to which Day 3 differentiated miPS-SMCs were transfected with Ex-mCherry or Ex- the transfection solution was added dropwise to the cells. After 16 h, the QKI-6 for 48 h. The cells were stimulated with 40 mM KCl for 15 min and medium was changed back to normal differentiation medium. Cells were live imaged using the Nikon 6D Inverted Microscope. Fluorescently harvested 48 h post-transfection. labelled cells (i.e. those successfully transfected) were measured for their cell size at time zero and 15 min. The percentage change in cell size was Reverse transcription of RNA and qRT-PCR calculated for the duration of the imaging. Total RNA was extracted using the RNeasy Mini Kit (Qiagen #74104) following the manufacturer’s protocol. 1–2 µg RNA was reverse transcribed In vitro tube formation into cDNA using the High Capacity cDNA Reverse Transcription Kit For the tube formation assays, 96-well plates were coated with growth-factor (ThermoFisher Scientific #4368814) with 10 µg of RNase inhibitor reduced Matrigel (Corning #354230). After defined treatments and time (ThermoFisher Scientific #N8080119) according to manufacturer’s protocol. points, differentiated cells were dissociated using TrypLE (Gibco #12604013) Thermocycler conditions were set at 25°C for 10 min, 37°C for 120 min, and counted, and 30,000 cells were seeded per well with 10 µM of the ROCK 85°C for 5 min and a 4°C hold. Resulting cDNA was diluted to a inhibitor Y27632 (TOCRIS #1254) in 150 µl medium. ECs and SMCs were concentration of 10 ng/µl with DEPC-H2O. seeded at a ratio of 2:1, respectively. Cells were incubated for between 4 and Relative gene expression was determined by qRT-PCR, using 20 ng of 8 h before imaging by bright field microscopy (DMi8, Leica). Images were cDNA per sample with the SYBR Green Master Mix (Life Technologies processed with the Angiogenesis Analyser plugin for ImageJ (Gilles #4368702) in a 10 μl reaction following the manufacturer’s protocol. For Carpentier Research Web Site: Computer Image Analysis). each sample 5 μl of PCR master mix, 2 μl of primer set, 1 μl DEPC-treated water and 2 μl of cDNA (20 ng) was used. Ct values were measured using In vivo Matrigel plug assay LightCycler 480 sequence detector (Roche). For each sample, PCR was Animals used in these studies were all bred in-house under constant climatic performed in duplicate in a 384-well reaction plate (LightCycler 480 Roche conditions with free access to food and water. All experiments were performed real-time PCR plates #04729749001) using primers specific to the gene in accordance with the Guidance on the Operation of the Animals (Scientific sequence (Table S1). The qPCR conditions were 5 min at 95°C for initial de- Procedures) Act, 1986 and approved by the Queen’s University Belfast Animal nature of the DNA followed by 40 cycles of 95°C for 15 s, 60°C for 30 s for Welfare and Ethical Review Body. Work was performed under the project the quantitative amplification stage with a final single cycle to create a license number PPL2821 and the Personal Licence Number 1705. melting curve with the conditions 95°C for 15 s, 60°C for 15 s then 95°C for On completion of cell differentiation and treatments, cells were 15 s. GAPDH served as the endogenous control to normalise the amounts of dissociated with TrypLE and counted using a haemocytometer. 1.5×105 RNA in each sample. The gene was considered undetectable beyond 35 miPS-ECs were combined with an equal number of miPS-SMCs per cycles. Primers are detailed in Table S1. treatment condition. The cells were centrifuged for 5 min at 1000 g to obtain the cell pellet. This was resuspended in 1 ml of medium and the cells were Immunoblotting stained with 5 µl Vybrant Dil Cell Labeling Solution (Thermo Fisher Cells were harvested and washed with cold PBS, resuspended in RIPA Scientific #V22885) per 1 ml of cell suspension, per 106 cells, for 15 min at buffer (Sigma #R0278) and lysed by ultra-sonication (Bradson 37°C. 2 ml of medium was then added to the suspension and the cells Sonifier150). The protein concentration in lysates was quantified with a centrifuged for 5 min at 200 g. The resulting pellet was resuspended in 50 µl

Quick Start Bradford protein assay (BioRad #5000201). Lysates were of PBS and layered on top of 200 µl Matrigel matrix basement membrane Journal of Cell Science

11 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

(Corning #356234). Three male 8–10-week-old C57BL6 mice were Bennett, M. R., Sinha, S. and Owens, G. K. (2016). Vascular smooth muscle cells anaesthetised using 1 litre min−1 of oxygen with 2% isoflurane (Attane, in Atherosclerosis. Circ. Res. 118, 692-702. doi:10.1161/CIRCRESAHA.115. #12164-002-25) for induction followed by 1.5% for maintenance of 306361 Black, D. L. (2003). Mechanisms of alternative pre-messenger RNA splicing. Annu. anaesthesia. Hair was removed from the lower back of the mouse and the Rev. Biochem. 72, 291-336. doi:10.1146/annurev.biochem.72.121801.161720 250 µl cell/Matrigel or PBS/Matrigel vehicle control suspension was Brinegar, A. E. and Cooper, T. A. (2016). Roles for RNA-binding proteins in injected subcutaneously. No more than 3 plugs were injected per mouse. development and disease. Brain Res. 1647, 1-8. doi:10.1016/j.brainres.2016.02. After 7 days, the mice were killed and the plugs were dissected from the 050 underside of the skin. Plugs were snap frozen in liquid nitrogen before Carmeliet, P. (2000). Mechanisms of angiogenesis and arteriogenesis. Nat. Med. 6, transfer into Tissue-Tek Optimal Cutting Temperature Compound (O.C.T.) 389-395. doi:10.1038/74651 − Chen, J., Kitchen, C. M., Streb, J. W. and Miano, J. M. (2002). Myocardin: a (VWR #25608-930) and storage at 80°C. Plugs were sectioned using the component of a molecular switch for smooth muscle differentiation. J. Mol. Cell. Leica CM1900 Cryostat into 12 µm slices and placed onto Superfrost Ultra Cardiol. 34, 1345-1356. doi:10.1006/jmcc.2002.2086 Plus Adhesion Slides (Thermo Fisher Scientific #J3800AMNT). Sections Clayton, Z. E., Sadeghipour, S. and Patel, S. (2015). Generating induced were allowed to reach room temperature for 30 min before freezing or pluripotent stem cell derived endothelial cells and induced endothelial cells for continuing with staining. cardiovascular disease modelling and therapeutic angiogenesis. Int. J. Cardiol. 197, 116-122. doi:10.1016/j.ijcard.2015.06.038 For Haematoxylin and Eosin staining, slides were hydrated for 5 min in Cochrane, A., Kelaini, S., Tsifaki, M., Bojdo, J., Vila-Gonzà ́lez, M., Drehmer, D., running distilled water to remove the O.C.T. compound. Slides were placed Caines, R., Magee, C., Eleftheriadou, M., Hu, Y. et al. (2017). Quaking is a key in haematoxylin stain for 5 min, washed for 3 min, submerged in acid regulator of endothelial cell differentiation, neovascularization, and angiogenesis. alcohol for 10 s then 1% ammonia for 30 s. A 3 min eosin submersion was Stem Cells 35, 952-966. doi:10.1002/stem.2594 carried out next, followed by a serial dehydration through two 75%, 95% Davies, P., Civelek, M., Fang, Y., Guerraty, M. and Passerini, A. (2010). and 100% alcohols for 1 min each. The slides were then cleared in Xylene Endothelial heterogeneity associated with regional athero-susceptibility and adaptation to disturbed blood flow in vivo. Semin. Thromb. Hemost. 36, three times for 3 min each. Slides where then mounted with DPX and when 265-275. doi:10.1055/s-0030-1253449 dry, imaged by bright field microscopy using a Leica DMi8 microscope. Debakey, M. E. and Glaeser, D. H. (2000). Patterns of atherosclerosis: effect of risk For immunofluorescence staining of frozen sections, slides were fixed factors on recurrence and survival—analysis of 11,890 cases with more than 25-year with cold acetone for 10 min before continuing from the permeabilisation follow-up. Am. J. Cardiol. 85, 1045-1053. doi:10.1016/S0002-9149(00)00694-9 step as detailed in the above fluorescence staining section. Images were Di Bernardini, E., Campagnolo, P., Margariti, A., Zampetaki, A., Karamariti, E., obtained using the a Leica DMi8 microscope. Hu, Y. and Xu, Q. (2014). Endothelial lineage differentiation from induced pluripotent stem cells is regulated by microRNA-21 and transforming growth factor β2 (TGF-β2) pathways. J. Biol. Chem. 289, 3383-3393. doi:10.1074/jbc.M113. Statistical analysis 495531 Data are expressed across independent biological replicates as the Dressel, U., Bailey, P. J., Wang, S.-C. M., Downes, M., Evans, R. M. and Muscat, mean±s.e.m. and were analysed using GraphPad Prism 5 software with a G. E. O. (2001). A dynamic role for HDAC7 in MEF2-mediated muscle two-tailed Student’s t-test for two groups or pairwise comparisons or differentiation. J. Biol. Chem. 276, 17007-17013. doi:10.1074/jbc.M101508200 P P P Ebersole, T. A., Chen, Q., Justice, M. J. and Artzt, K. (1996). The quaking gene analysis of variance (ANOVA). * <0.05; ** <0.01; *** <0.001 was product necessary in embryogenesis and myelination combines features of RNA considered significant. binding and signal transduction proteins. Nat. Genet. 12, 260-265. doi:10.1038/ ng0396-260 Competing interests Fagg, W. S., Liu, N., Fair, J. H., Shiue, L., Katzman, S., Donohue, J. P. and Ares, The authors declare no competing or financial interests. M. (2017). Autogenous cross-regulation of Quaking mRNA processing and translation balances Quaking functions in splicing and translation. Genes Dev. 31, 1894-1909. doi:10.1101/gad.302059.117 Author contributions Fu, X.-D. and Ares, M. (2014). Context-dependent control of alternative splicing by Conceptualization: R.C., L.Z., A. Margariti; Methodology: R.C., A.C., S.K., M.V., RNA-binding proteins. Nat. Rev. Genet. 15, 689-701. doi:10.1038/nrg3778 C.Y., M.E., A. Moez, D.G., A. Margariti; Validation: R.C., A. Margariti; Formal Galarneau, A. and Richard, S. (2005). Target RNA motif and target mRNAs of the analysis: R.C., A.C., S.K., A. Margariti; Investigation: R.C., A.C., S.K., M.V., C.Y., Quaking STAR protein. Nat. Struct. Mol. Biol. 12, 691-698. doi:10.1038/nsmb963 M.E., A. Moez, A. Margariti; Resources: A.S., D.G., A. Margariti; Writing - original Gomez, D. and Owens, G. K. (2012). Smooth muscle cell phenotypic switching in draft: R.C.; Writing - review & editing: R.C., A.S., L.Z., D.G., A. Margariti; atherosclerosis. Cardiovasc. Res. 95, 156-164. doi:10.1093/cvr/cvs115 Supervision: D.G., A. Margariti; Project administration: A. Margariti; Funding Guallar, D. and Wang, J. (2014). RNA-binding proteins in pluripotency, acquisition: A. Margariti. differentiation, and reprogramming. Front Biol9, 389-409. doi:10.1007/s11515- 014-1326-y Funding Hall, M. P., Nagel, R. J., Fagg, W. S., Shiue, L., Cline, M. S., Perriman, R. J., This work was supported by grants from the British Heart Foundation (FS/15/23/ Donohue, J. P. and Ares, M. (2013). Quaking and PTB control overlapping 31435, PG/16/8/31985 and PG/18/29/33731) and Biotechnology and Biological splicing regulatory networks during muscle cell differentiation. RNA 19, 627-638. Sciences Research Council (BBSRC) (BB/M003221/1). doi:10.1261/rna.038422.113 Hao, H., Ropraz, P., Verin, V., Camenzind, E., Geinoz, A., Pepper, M. S., Gabbiani, G. and Bochaton-Piallat, M.-L. (2002). Heterogeneity of smooth Supplementary information muscle cell populations cultured from pig coronary artery. Arterioscler. Thromb. Supplementary information available online at Vasc. Biol. 22, 1093-1099. doi:10.1161/01.ATV.0000022407.91111.E4 http://jcs.biologists.org/lookup/doi/10.1242/jcs.230276.supplemental Kelaini, S., Cochrane, A. and Margariti, A. (2014). Direct reprogramming of adult cells: Avoiding the pluripotent state. Stem Cells Cloning Adv Appl. 7, 19-29. References doi:10.2147/SCCAA.S38006 Alfonso, F., Byrne, R. A., Rivero, F. and Kastrati, A. (2014). Current treatment of in- Kim, D.-Y., Kim, K.-M., Kim, E.-J. and Jang, W.-G. (2018). Hypothermia-induced stent restenosis. J. Am. Coll. Cardiol. 63, 2659-2673. doi:10.1016/j.jacc.2014.02.545 RNA-binding motif protein 3 (RBM3) stimulates osteoblast differentiation via the Baralle, F. E. and Giudice, J. (2017). Alternative splicing as a regulator of ERK signaling pathway. Biochem. Biophys. Res. Commun. 498, 459-465. doi:10. development and tissue identity. Nat. Rev. Mol. Cell Biol. 18, 437-451. doi:10. 1016/j.bbrc.2018.02.209 1038/nrm.2017.27 Kondo, T., Furuta, T., Mitsunaga, K., Ebersole, T. A., Shichiri, M., Wu, J., Artzt, Belaguli, N. S., Zhou, W., Trinh, T.-H. T., Majesky, M. W. and Schwartz, R. J. K. and Abe, K. (1999). Genomic organization and expression analysis of the mouse qkI locus. Mamm. Genome 10, 662-669. doi:10.1007/s003359901068 (1999). Dominant negative murine serum response factor: alternative splicing Li, Z., Takakura, N., Oike, Y., Imanaka, T., Araki, K., Suda, T., Kaname, T., within the activation domain inhibits transactivation of serum response factor Kondo, T., Abe, K. and Yamamura, K. (2003). Defective smooth muscle binding targets. Mol. Cell. Biol. 19, 4582-4591. doi:10.1128/MCB.19.7.4582 development in qkI-deficient mice. Dev. Growth Differ. 45, 449-462. doi:10.1111/j. Beltramo, E. and Porta, M. (2013). Pericyte loss in diabetic retinopathy: 1440-169X.2003.00712.x mechanisms and consequences. Curr. Med. Chem. 20, 3218-3225. doi:10. Libby, P., Ridker, P. M. and Hansson, G. K. (2011). Progress and challenges in 2174/09298673113209990022 translating the biology of atherosclerosis. Nature 473, 317-325. doi:10.1038/ Benjamin, L. E., Hemo, I. and Keshet, E. (1998). A plasticity window for blood nature10146 vessel remodelling is defined by pericyte coverage of the preformed endothelial Lilly, B. (2014). We have contact: endothelial cell-smooth muscle cell interactions.

network and is regulated by PDGF-B and VEGF. Development 125, 1591-1598. Physiology 29, 234-241. doi:10.1152/physiol.00047.2013 Journal of Cell Science

12 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs230276. doi:10.1242/jcs.230276

Majesky, M. W. (2007). Developmental basis of vascular smooth muscle diversity. Takahashi, K. and Yamanaka, S. (2006). Induction of pluripotent stem cells from Arterioscler. Thromb. Vasc. Biol. 27, 1248-1258. doi:10.1161/ATVBAHA.107. mouse embryonic and adult fibroblast cultures by defined factors. Cell 126, 141069 663-676. doi:10.1016/j.cell.2006.07.024 Mandler, M. D., Ku, L. and Feng, Y. (2014). A cytoplasmic quaking I isoform Van Der Veer, E. P., De Bruin, R. G., Kraaijeveld, A. O., De Vries, M. R., Bot, I., regulates the hnRNP F/H-dependent alternative splicing pathway in myelinating Pera, T., Segers, F. M., Trompet, S., Van Gils, J. M., Roeten, M. K. et al. (2013). glia. Nucleic Acids Res. 42, 7319-7329. doi:10.1093/nar/gku353 Quaking, an RNA-binding protein, is a critical regulator of vascular smooth muscle Margariti, A., Xiao, Q., Zampetaki, A., Zhang, Z., Li, H., Martin, D., Hu, Y., Zeng, cell phenotype. Circ. Res. 113, 1065-1075. doi:10.1161/CIRCRESAHA.113. L. and Xu, Q. (2009). Splicing of HDAC7 modulates the SRF-myocardin complex 301302 during stem-cell differentiation towards smooth muscle cells. J. Cell Sci. 122, Wang, Z., Wang, D.-Z., Hockemeyer, D., Mcanally, J., Nordheim, A. and Olson, 460-470. doi:10.1242/jcs.034850 E. N. (2004). Myocardin and ternary complex factors compete for SRF to control Margariti, A., Winkler, B., Karamariti, E., Zampetaki, A., Tsai, T.-N., Baban, D., Ragoussis, J., Huang, Y., Han, J.-D. J., Zeng, L. et al. (2012). Direct smooth muscle gene expression. Nature 428, 185-189. doi:10.1038/nature02382 reprogramming of fibroblasts into endothelial cells capable of angiogenesis and Wang, E. T., Sandberg, R., Luo, S., Khrebtukova, I., Zhang, L., Mayr, C., reendothelialization in tissue-engineered vessels. Proc. Natl. Acad. Sci. USA 109, Kingsmore, S. F., Schroth, G. P. and Burge, C. B. (2008). Alternative isoform 13793-13798. doi:10.1073/pnas.1205526109 regulation in human tissue transcriptomes. Nature 456, 470-476. doi:10.1038/ Modrek, B., Resch, A., Grasso, C. and Lee, C. (2001). Genome-wide detection of nature07509 alternative splicing in expressed sequences of human genes. Nucleic Acids Res. Wang, G., Xiao, Q., Luo, Z., Ye, S. and Xu, Q. (2012). Functional impact of 29, 2850. doi:10.1093/nar/29.13.2850 heterogeneous nuclear ribonucleoprotein A2/B1 in smooth muscle differentiation Moiseeva, E. P. (2001). Adhesion receptors of vascular smooth muscle cells and from stem cells and embryonic arteriogenesis. J. Biol. Chem. 287, 2896-2906. their functions. Cardiovasc. Res. 52, 372-386. doi:10.1016/S0008- doi:10.1074/jbc.M111.297028 6363(01)00399-6 Xie, N., Chen, M., Dai, R., Zhang, Y., Zhao, H., Song, Z., Zhang, L., Li, Z., Feng, Y., Montero, D., Walther, G., Pérez-Martin, A., Vicente-Salar, N., Roche, E. and Gao, H. et al. (2017). SRSF1 promotes vascular smooth muscle cell proliferation Vinet, A. (2013). Vascular smooth muscle function in type 2 diabetes mellitus: a through a Δ133p53/EGR1/KLF5 pathway. Nat. Commun. 8, 16016. doi:10.1038/ systematic review and meta-analysis. Diabetologia 56, 2122-2133. doi:10.1007/ ncomms16016 s00125-013-2974-1 Xu, Y., Li, R., Zhang, K., Wu, W., Wang, S., Zhang, P. and Xu, H. (2018). The Neff, L. P., Tillman, B. W., Yazdani, S. K., Machingal, M. A., Yoo, J. J., Soker, S., multifunctional RNA-binding protein hnRNPK is critical for the proliferation and Bernish, B. W., Geary, R. L. and Christ, G. J. (2011). Vascular smooth muscle differentiation of myoblasts. BMB Rep. 51, 350-355. doi:10.5483/BMBRep.2018. enhances functionality of tissue-engineered blood vessels in vivo. J. Vasc. Surg. 53, 426-434. doi:10.1016/j.jvs.2010.07.054 51.7.043 Nilsen, T. W. and Graveley, B. R. (2010). Expansion of the eukaryotic proteome by Yabas, M., Elliott, H., Hoyne, G., Yabas, M., Elliott, H. and Hoyne, G. F. (2015). alternative splicing. Nature 463, 457-463. doi:10.1038/nature08909 The role of alternative splicing in the control of immune homeostasis and cellular Noveroske, J. K., Lai, L., Gaussin, V., Northrop, J. L., Nakamura, H., Hirschi, differentiation. Int. J. Mol. Sci. 17, 3. doi:10.3390/ijms17010003 K. K. and Justice, M. J. (2002). Quaking is essential for blood vessel Yamashita, J., Itoh, H., Hirashima, M., Ogawa, M., Nishikawa, S., Yurugi, T., development. Genesis 32, 218-230. doi:10.1002/gene.10060 Naito, M., Nakao, K. and Nishikawa, S.-I. (2000). Flk1-positive cells derived from Owens, G. K., Kumar, M. S. and Wamhoff, B. R. (2004). Molecular regulation of embryonic stem cells serve as vascular progenitors. Nature 408, 92-96. doi:10. vascular smooth muscle cell differentiation in development and disease. Physiol. 1038/35040568 Rev. 84, 767-801. doi:10.1152/physrev.00041.2003 Yang, J., Margariti, A. and Zeng, L. (2016). Analysis of Histone Deacetylase 7 Park, K.-H. and Park, W. J. (2015). Endothelial dysfunction: clinical implications in (HDAC7) alternative splicing and its role in embryonic stem cell differentiation cardiovascular disease and therapeutic approaches. J. Korean Med. Sci. 30, toward smooth muscle lineage. Methods Mol. Biol. 1436, 95-108. doi:10.1007/ 1213-1225. doi:10.3346/jkms.2015.30.9.1213 978-1-4939-3667-0_7 Paz, I., Kosti, I., Ares, M., Jr, Cline, M. and Mandel-Gutfreund, Y. (2014). Yang, C., Kelaini, S., Caines, R. and Margariti, A. (2018). RBPs play important RBPmap: a web server for mapping binding sites of RNA-binding proteins. roles in vascular endothelial dysfunction under diabetic conditions. Front. Physiol. Nucleic Acids Res. 42, W361-W367. doi:10.1093/nar/gku406 9, 1310. doi:10.3389/fphys.2018.01310 Qian, H., Yang, Y., Li, J., Huang, J., Dou, K. and Yang, G. (2007). The role of Zhang, L., Jin, M., Margariti, A., Wang, G., Luo, Z., Zampetaki, A., Zeng, L., Ye, vascular stem cells in atherogenesis and post-angioplasty restenosis. Ageing S., Zhu, J. and Xiao, Q. (2010). Sp1-dependent activation of HDAC7 is required Res. Rev. 6, 109-127. doi:10.1016/j.arr.2007.01.001 Roth, G. A., Huffman, M. D., Moran, A. E., Feigin, V., Mensah, G. A., Naghavi, M. for platelet-derived growth factor-BB-induced smooth muscle cell differentiation and Murray, C.J. (2015). Global and regional patterns in cardiovascular mortality from stem cells. J. Biol. Chem. 285, 38463-38472. doi:10.1074/jbc.M110.153999 from 1990 to 2013. Circulation 132, 1667-1678. doi:10.1161/CIRCULATIONAHA. Zhou, B., Margariti, A., Zeng, L., Habi, O., Xiao, Q., Martin, D., Wang, G., Hu, Y., 114.008720 Wang, X. and Xu, Q. (2011). Splicing of histone deacetylase 7 modulates smooth Schatteman, G. C., Loushin, C., Li, T. and Hart, C. E. (1996). PDGF-A is required muscle cell proliferation and neointima formation through nuclear β-catenin for normal murine cardiovascular development. Dev. Biol. 176, 133-142. doi:10. translocation. Arterioscler. Thromb. Vasc. Biol. 31, 2676-2684. doi:10.1161/ 1006/dbio.1996.9988 ATVBAHA.111.230888 Journal of Cell Science